1

Manuscriptdate:26October2015HowtoNeutralizeaFiniteClauseBoundary:PhaseTheoryandtheGrammarofBoundPronounsThomasGranoHowardLasnikAbstract: A bound pronoun in the subject position of a finite complement clauserenders theclauseboundaryrelatively transparent toprocessesordinarily limitedtomonoclausal,control,andraisingconfigurations.Wearguethatthisobservationsupports a convergence-based approach to phasehood, as well as the view thatbound pronouns sometimes enter the derivation with unvalued phi-features. Wesketch various ways of implementing this core idea, and draw connections tocontrol,fakeindexicals,andislandeffects.Keywords: bound pronouns, phase theory, clause-mate conditions, control, fakeindexicals,islands1IntroductionThis paper has two main goals. The first goal is to lay out data on an under-documented phenomenon in English syntax whereby a bound pronoun in thesubject position of a finite complement clause renders the clause boundaryrelativelytransparenttoprocessesorrelationsordinarilyconfinedtomonoclausal,control,andraisingconfigurations.(ThisobservationandmanyoftherelevantfactsarelaidoutbyLasnik2006.Seealsosection2belowforotherpertinentreferences.)Thesecondgoalistosuggestanaccountofthisphenomenonthathasrepercussionsfortwoareasofgrammar,namelyphasetheoryandboundpronouns.

Thebasicobservationisillustratedin(1)-(2)forgapping,andwewillshowbelowthatthesamepatternholdsforawiderangeofothersyntacticphenomenaaswell.(1a)illustratesabaselineunremarkableexampleofgapping,and(1b)and(1c)showthatthegappedmaterialcanspanaraisingorcontrolboundaryrespectively.(2a) shows thatgappingordinarily cannot spana finite clauseboundary,but (2b)shows that this restriction is loosened if the subject of the embedded clause is aboundpronoun.Wecallthisthe“boundpronouneffect”.Ourconcerninthispaperis why the presence of the bound pronoun renders sentences like (2b) so muchbetterthanexpected;thefactthatsuchsentencesarenotquiteasacceptableastheoneswithinfinitivalcomplementsweleavetofutureresearch.

(1) a.JoelikesapplesandTim<likes>oranges.

b.Joe1seemst1tolikeapplesandTim2<seemst2tolike>oranges.c.Joe1claimsPRO1tolikeapplesandTim2<claimsPRO2tolike>oranges.

2

(2) a.*Joe1claimsthatBilllikesapplesandTim2<claimsthatBilllikes>oranges.b.?Joe1claimsthathe1likesapplesandTim2<claimsthathe2likes>oranges.

The bound pronoun effect invites a very simple analysis, namely thatsketchedin(3).(3) a.Gappingisphase-bound. b.Boundpronounsenterthederivationwithunvaluedphi-features. c.Unvaluedfeaturesvoidphasehood.If the locality domain for gapping is the phase, and bound pronouns enter thederivationinawaythatvoidstheotherwisephasalstatusofthecomplementclause,thenthecontrastbetween(2a)and(2b)follows. Butawiderrangeofdatarevealsamorenuancedpicture.Inparticular,theboundpronoun effect holds onlywhen it is the subject of the complement clausethat is theboundpronoun.As illustrated in (4a), non-subject boundpronounsdonot improve acceptability. Furthermore, the bound pronoun must be the entiresubject of the complement clause: as illustrated in (4b), subject-internal boundpossessorsdonothavetheamelioratingeffect.(4) a. *Joe1 claims that Bill gave him1 apples and Tim2 <claims that Bill gave

him2>oranges.b.*Joe1claimsthathis1sonlikesapplesandTim2<claimsthathis2sonlikes>oranges.

Rather than give up on the account in (3) altogether, though, we think that theadditional data warrant a qualified version of (3), whereby bound pronounssometimes but not always enter the derivation with unvalued phi-features, and(possibly)whereby unvalued features sometimes but not always void phasehood.Our task for the rest of this paper is to sketch out ways of making thesemodifications precise, as well as to discuss their implications for the analysis ofislandsandfortheanalysisofotherphenomenaforwhichunvaluedpronounshavebeeninvoked,namelycontrolandfakeindexicals. Therestofthepaperisorganizedasfollows.Section2summarizessomeofthepreviousliteratureonclause-mateconditionsandlaysoutdataillustratingtheboundpronouneffectforawiderangeofsyntacticprocessesandrelations.Section3developsaphase-theoreticaccountofthedataandcomparestwoalternativewaysofimplementingthebasicproposal.Section4discussesinterventioneffects.Section5discussesconsequencesforcontrolandfakeindexicals,andsection6bringsislandeffectsintothepicture.Finally,section7concludes.2TheEmpiricalLandscapeandSomePreviousLiteratureTheideathatsomesyntacticprocessesandrelationscannotcrossaclauseboundaryhasplayedaroleingenerativetheorizingsincethe1950s.(SeeLasnik2002foranoverview.)Ithasalsolongbeenobservedthatnotallclauseboundariesarecreated

3

equal: Chomsky’s (1973) Tensed Sentence Condition and Specified SubjectConditionbothacknowledge therelativeweaknessofnonfiniteclauseboundaries.Inarelatedvein,Postal(1974)usestheterm“quasi-clause”(acoinageheattributesto Perlmutter) for raising and control complements, and he suggests that quasi-clauseboundariesare“notasstrongabarriertoatleastsomesyntacticphenomenaas full clause boundaries” (p. 232). Postal invokes quasi-clauses in discussing arangeofprocessesandrelationsthatappeartobeabletospananonfinitebutnotafinite clause boundary, including heavy NP shift, comparative deletion, toughmovement,1multiple questions, and double negation (the latter obeying an “anti-clause-mate” condition). (The clause-boundedness of multiple questions is notedalsobyKuno&Robinson1973.) Other phenomena for which clause-boundedness and the finite/nonfinitedistinctionhavesincebeenfoundtoberelevant includeinversescope(May1985;Larson&May1990;Hornstein1994;Farkas&Giannakidou1996;Kennedy1997;Kayne 1998; Fox 2000; Moulton 2008; Wurmbrand 2011), antecedent-containeddeletion (Larson &May 1990; Hornstein 1994; Kennedy 1997), multiple sluicing(Merchant2001;Nishigauchi 1998;Barrie 2008; Lasnik2014), reciprocal binding(Higginbotham 1981), “family of questions” readings (May 1977, 1985; Williams1986;Sloan&Uriagereka1988; Sloan1991;Lasnik2006;Agüero-Bautista2007),andslangNPIlicensing(Lasnik2002).

But for some of these phenomena, not all authors claim that the relevantdistinctionissimplybetweenfiniteclausesandnonfiniteclauses.OnetrendintheliteraturebuildsonRizzi’s(1978)seminalworkonItalian,wherecliticclimbingandrelatedphenomenaareshowntobeordinarilyclause-boundexceptinsomebutnotallsentencesinvolvingnonfinitecomplementation.Crucially,Rizzishowedthattheavailability of clitic climbing across a nonfinite clauseboundary is conditionedbythe choice of the embedding verb, generalizing that only modal, motion, andaspectualverbsextendlocality.Thereisnowasizeable literatureonrestructuringthatcorroboratesversionsof thisclaimforanalogouseffects inSpanish(Aissen&Perlmutter 1983), German (Wurmbrand 2001; Lee-Schoenfeld 2007), andpotentially a much wider range of languages as well (Cinque 2004; Grano 2015;Wurmbrand2015).

In thisconnection,Lechner(2001),buildingon Johnson(1996),claimsthatgapping and comparative deletion in English only apply across nonfinite clauseboundariesiftheembeddingverbisarestructuringverb.Hornstein(1994)makes1Chomsky(1981:314)claimsthattoughmovementacrossafiniteclauseboundaryis“sometimesmoreorlessacceptable”,citingtheexamplein(i).

(i) Thisbookisdifficulttoconvincepeoplethattheyoughttoread.Notably,thesubjectoftherelevantclauseinthisexampleisaboundpronoun.Butunlikemostoftheexamplesoftheboundpronouneffectconsideredinthispaper,thepronounisboundbytheobjectratherthanthesubjectoftheembeddingclause.Formoreontheissueofobjectbinders,seethediscussionofinterventioninsection4below.

4

the same claim for inverse scope and antecedent-contained deletion (though thisviewisquestionedbyKennedy1997andWurmbrand2013;seealsoMoulton2008forrelevantexperimentalworkoninversescope).Anotherphenomenonforwhichrestructuring has been invoked in English is infinitival to contraction in locutionslikewanna(fromwantto)orhafta(fromhaveto):seeGoodall1991.Inadifferentvein, Grano (2012, 2015) observes that in English, a control verb’s (in)ability toembeda finitecomplement(aswellasotherrelatedproperties)closelytrackstheverb’sstatusasa(non-)restructuringverbinlanguageslikeGermanandItalian.

Theboundpronouneffectinvestigatedinthispaperconstitutesyetanotherchallenge to the characterization of locality domains in terms of a simplefinite/nonfinitecontrast: justastherestructuringliteraturehasshownthatnotallnonfinitecomplementsarecreatedequal,theboundpronouneffectsshowsthatnotallfinitecomplementsarecreatedequaleither.Theboundpronouneffecthasbeenobserved before, by Sloan (1991) for “family of questions” readings, Nishigauchi(1998) for multiple sluicing in Japanese, and Merchant 2001:113 note 4 (citingNishigauchi)formultiplesluicingaswellasgapping.(SeealsoLasnik2014.)Lasnik(2006)assemblesarangeofdata fromEnglishshowing theboundpronouneffectfor gapping, pseudogapping, inverse scope, antecedent-contained deletion,reciprocal binding, heavy NP shift, multiple sluicing, and “family of questions”readings.Inapossiblyrelatedvein,Ruys(1992)andKratzer(1998b)observethatboundpronounsfacilitatetheavailabilityof intermediatescopeincertainkindsofsyntacticconfigurations.2

The starting point for this paper is that for a wide range of “clause-mate”phenomena, nonfinite complement clauses pattern like finite complement clauseswithboundpronominalsubjectsinbeingfullyacceptable(intheformercase)oratleast somewhatacceptable (in the lattercase),whereas finitecomplementclauseswithout bound pronominal subjects give rise to categorical unacceptability. Thedatasupportingthispositionaregivenasaseriesofminimaltriosin(5)-(14),eachtrioexemplifyingadifferentclause-matephenomenon.Ineachcase,thematrixverbis claim. We use claim for two reasons. First, claim readily accepts both finitecomplements and controllednonfinite complementswith little ornodifference inmeaning,thusfacilitatingtheformationofminimaltrios.Second,claimbelongstoaclassofverbsthatresistrestructuringcross-linguistically(Wurmbrand2001),andso we take the acceptability of the (a) sentences below as evidence that theseclause-mate phenomena are not sensitive to the cross-linguistically attested splitbetweenrestructuringandnon-restructuringverbs.

2Aside from restructuring and the bound pronoun effect, other potentialcomplications for the characterization of locality domains include control/raisingasymmetries (there is agreement that inverse scope is possible out of controlcomplements but disagreement about whether it is possible out of raisingcomplements: Wurmbrand 2013; Frank & Storoshenko 2015) and asymmetriesbetweencontrolandraisingcomplementsontheonehandandECMorraising-to-objectcomplementsontheotherhand.

5

(5) GAPPINGa.Joe1claimsPRO1tolikeapplesandTim2<claimsPRO2tolike>oranges.b.?Joe1claimsthathe1likesapplesandTim2<claimsthathe2likes>oranges.c.*Joe1claimsthatBilllikesapplesandTim2<claimsthatBilllikes>oranges.

(6) PSEUDOGAPPINGa. Joe1claimsPRO1tolikeapplesandbuthedoesn’t<claimPRO2tolike>

oranges.b. ?Joe1claimsthathe1likesapplesbuthedoesn’t<claimthathe2likes>

oranges.c. *Joe1claims thatBill likesapplesandbuthedoesn’t<claim thatBill

likes>oranges.(7) COMPARATIVEDELETION

a. More people1 claim PRO1 to like apples than <claim PRO1 to like>oranges.

b. ?More people1 claim that they1 likes apples than <claim that they1like>oranges.

c. *Morepeople1claimthatBill likesapplesthan<claimthatBill likes>oranges.

(8) INVERSESCOPE a.Atleastoneprofessorclaimstoreadeveryjournal. ∃>∀ b.Atleastoneprofessor1claimsthathe1readseveryjournal. ? ∃>∀ c.Atleastoneprofessor1claimsthatBillreadseveryjournal. *∃>∀(9) ANTECEDENT-CONTAINEDDELETION a.Joe1claimsPRO1toreadeverythingTim2does<claimPRO2toread>. b.?Joe1claimsthathe1readseverythingTim2does<claimthathe2reads>. c.*Joe1claimsthatBillreadseverythingTim2does<claimthatBillreads>.(10) MULTIPLEQUESTIONS a.Tellmewho1claimsPROtoreadwhichjournal. b.?Tellmewho1claimsthatthey1readwhichjournal. c.*Tellmewho1claimsthatBillreadswhichjournal.(11) TOUGHMOVEMENT a.ThismagazineistoolowbrowforJohn1toclaimPRO1toread. b.?ThismagazineistoolowbrowforJohn1toclaimthathe1reads. c.*ThismagazineistoolowbrowforJohn1toclaimthatBillreads.(12) RECIPROCALBINDING a.[JohnandBill]1claimPRO1tolikeeachother. Intendedreading:JohnclaimstolikeBillandBillclaimstolikeJohn. b.?[JohnandBill]1claimthatthey1likeeachother.

6

Intendedreading:JohnclaimsthathelikesBillandBillclaimsthathelikesJohn.

c.*[JohnandBill]1claimsthatTimlikeseachother.Intendedreading: John claims thatTim likesBill andBill claims thatTimlikesJohn.

(13) MULTIPLESLUICING

a. Someone1 claims PRO1 to beworried about something, but I don’twho2<claimsPRO2tobeworried>aboutwhat.b.?Someone1claimsthatthey1‘reworriedaboutsomething,butIdon’twho2<claimsthatthey2‘reworried>aboutwhat.c. *Someone1 claims that Bill’s worried about something, but I don’t who2<claimsthatBill’sworried>aboutwhat.

(14) FAMILYOFQUESTIONS a.Whichjournaldoeseveryone1claimPRO1toread?

Anticipated answer type: John claims to read LI, Tim claims to readNLLT,etc.

b.?Whichjournaldoeseveryone1claimthatthey1read?Anticipatedanswertype:JohnclaimsthathereadsLI,TimclaimsthathereadsNLLT,etc.

c.*Whichjournaldoeseveryone1claimthatBillreads?Anticipatedanswertype:JohnclaimsthatBillreadsLI,TimclaimsthatBillreadsNLLT,etc.

Inwhatfollows,weexplorewaysofmodelingagrammarthattreatsthe(a)and(b)sentences above as grammatical to the exclusion of the (c) sentences. While wethink this basic cut is right, the somewhat degraded acceptability of the (b)sentencesincomparisonwiththe(a)sentencesisunfortunatelynotsomethingthatwewilladdressinthispaper.3TowardanAnalysis3.1PreliminariesWebeginbytakingitasaplausibleworkinghypothesisthatthelocalitydomainforgappingandtheotherphenomenacataloguedaboveisthephase.Then,thecontrastingrammaticalitybetween(2a)and(2b)(repeatedherein(15))canbeaccountedforbythepairofproposalsin(16),whatwewillcalltheproto-analysis.(15) a.*Joe1claimsthatBilllikesapplesandTim2<claimsthatBilllikes>oranges.

b.?Joe1claimsthathe1likesapplesandTim2<claimsthathe2likes>oranges.(16)Proto-analysis

a. Convergence:Any unvalued feature anywhere in a candidate phase keepsthephaseopen.

7

b. Transmission: A pronoun with unvalued phi-features can be valued viafeaturetransmissionfromac-commandingbinderatanarbitrarydistance.

According to (16), the grammaticality of (15b) follows from the presence of theboundpronouninthecomplementclause.Theboundpronounentersthederivationwithunvaluedphi-featuresthatarenotvalueduntilthematrixbinderismergedin.ThepersistenceoftheunvaluedfeaturesthroughthebuildingofthecomplementCPprevents that CP from being a phase, thereby allowing gapping across the finiteclause boundary. In (15a), on the other hand, there is no bound pronoun in thecomplementclause,sotheembeddedCPisphasalandcannotparticipateingappingwiththematrixclause.

Butasalreadypreviewedabove, facts like (4), repeatedhere in (17), showthat the proto-analysis as it stands overgenerates: (17a) shows that a boundpronoun in object position does not extend the domain for gapping, and (17b)showsthatasubject-internalboundpossessordoesnoteither.(17) a.*Joe1 claims that Bill gave him1 apples and Tim2 <claims that Bill gave

him2>oranges.b.*Joe1claimsthathis1sonlikesapplesandTim2<claimsthathis2sonlikes>oranges.

WetakethesefactstomeanthateitherConvergenceneedstobereinedinsothatunvalued features keep the phase open only under some conditions, orTransmission needs to be reined in so that a binder can transmit features to anunvaluedpronounonlywhenthebinderandthepronounstandinaparticularkindofconfigurationalrelationshipwitheachother.Sections3.2and3.3belowdiscusseachofthesetwopossibilitiesinturn.First,though,weelaborateinabitmoredetailonsomeofthenuancesinvolvedinunderstandingConvergenceandTransmission,respectively.3.1.1ConvergenceThe convergence approach to phasehood is entertained by Chomsky (2000:107),butheultimatelyfavorsthealternativeviewthatphasesaredefinedasvPandCP,on the grounds that the convergence viewwould require look-ahead.3Chomsky’s3WhyvPandCP?Dotheyconstituteanaturalclass?Chomsky(2000:106)suggeststhatphasesshouldbe“relativelyindependentintermsofinterfaceproperties”.Hegoesontosuggestthatsemantically,vPandCPare“theclosestsyntacticcounterparttoaproposition:eitheraverbphraseinwhichalltheta-rolesareassignedorafullclauseincludingtenseandforce”andthatphonologically,vPandCPsharecertainpropertiessuchas“fronting,extraposition,pseudoclefting,responsefragments,etc.”(p.106).Itisnotclear,though,howtodefine“proposition”insuchawaythatitpicksoutvPandCPtotheexclusionofTP,andsothisleavesuswitha“listproblem”:thesetofphasetypeshastobestipulatedratherthanfollowingfromsomethingmoregeneral.Itisinterestingtonotethatversionsofthe“listproblem”

8

argument goes as follows. The contrast in acceptability between (18a) and (18b)suggestsaMergeoverMovePrinciple:whatgoeswrongin(18b)isthatatthepointinthederivationwhereembedded[Spec,TP]isbuilt,theDPmanylinguistsmovestofillitratherthantheexpletivetherebeingmergedinfromthelexicalarray.Buttheninorder to explainwhy there isnoviolationof theMergeoverMovePrinciple in(19),we can say thatMerge overMove applies over lexical subarrays,which areorganizedintophases.SincethereisinahigherphaseduetotheinterveningCP,itisnoteligibletobemerged in,soMovedoesnotcompetewith it.But if this isright,then(20)isunderivableonaconvergence-basedapproachtophasehood,providedthat wh-phrases have uninterpretable features: the uninterpretable feature onwhich should void the phase status of the embedded CP. Consequently, given theMergeoverMovePrinciple,thereshouldbeforcedtomergeinat[Spec,TP]ofwill.On the view that phases are defined as vPandCP, on the other hand, there is noproblemderiving(20),sincetheembeddedCPisphasaldespiteitsuninterpretiblefeature, thereby preventing a situation wherein the Merge over Move Principleappliesandforcesprematuremergingofthere. (18) a.Therearelikely[TPtheretobemanylinguistsatthisconference].

b. *There are likely [TP many linguists to be many linguists at thisconference].

(19) Thereissomelikelihood[CPthat[TPmanylinguistswillbemanylinguistsat

thisconference]].(20) Whichconferenceistheresomelikelihood[CPthatmanylinguistswillbeat

whichconference]? Felser (2004), however, in her study ofwh-copying, argues in favor of theconvergence approach based on the following reasoning: (1) wh-phrases haveuninterpretablefeaturesthatpersistuntil thewh-phraseis in its final landingsite,(2) wh-phrases move successive-cyclically through intermediate [Spec,CP]positions, and (3) a constituent with uninterpretable features cannot be sent toSpellout.Takentogether,theseconsiderationsimplythattheintermediateCPsinamulticlausalwh-configurationcannotbephases,whichwouldfollowiftheoffendingintermediate wh-copies are in fact what void phasehood due to theiruninterpretableorunvaluedfeatures.(Whatisstillleftunsolvedonthisapproach,Felseracknowledges,iswhattriggersmovementtointermediatenon-interrogativeheadsinthefirstplace.Inresponsetothis,FelserendsupsuggestingadistinctionbetweenLFphases,whicharedefinedoverconvergence,andPFphases,whichformarefoundelsewhereinChomsky’swork;forexample,Chomsky’s(1973)TensedSentenceConditionandSpecifiedSubjectConditionarebothsubsumedunderthenotionofGovernmentinChomsky1981,butburiedinthedefinitionofGoverningCategoryisthetermofartSUBJECT(allcaps),whichChomskydefineswithalist:finiteAGR(supplantingtheTensedSentenceCondition)andthesubjectofanonfiniteclause(supplantingtheSpecifiedSubjectCondition).

9

“relatively independent phonological or processing units” (p. 570). Then,intermediatewh-movement is forced so as not to impede PF-Spellout.) To defuseChomsky’sargumentagainstconvergenceassketched in theprecedingparagraph,Felsersuggeststhatthereisnotatrueexpletive,sothatitsmergesiteisconstrainedthematically by the choice of the predicate and it does not actually competewithmovement. A question raised by the convergence approach to phasehood is how thePhase Impenetrability Condition would have to be formulated (or alternatively,what would replace it). According to Chomsky (2000, 2001), the reason phasesinduceopacityisbecauseatsomespecifiedpointinthederivation,thecomplementto the phase head is sent to Spellout and consequently cannot participate insubsequentsyntacticprocesses.TheworkingsofthisareclearifphaseheadsareaprioridefinedasC,v, etc.,but ifphasesaredefinedbyconvergence, it isnotclearwhat would constitute a phase head. One possible solution — implicit in ourproposal in (16a) above — is to entertain a hybrid approach to phasehooddeterminationwherein C and v are candidate phase headswhose status as phaseheads isvoided if their complementhasoneormoreunvalued features in it.Thispossibility is in fact entertained by Felser (2004).Wurmbrand (2011) suggests asimilar idea, proposing that “Only interpretationally complete units can betransferred…iF:___inapotentialphaseprojectionpostponestransfer”(p.69).JuanUriagareka (p.c.) has similarly offered the suggestion that “transfer is suspendedwhenananaphoricdependencyisatstake(untiltheantecedententersthepicture)”,an idea whose roots are also apparent in Uriagareka & Lasnik 2005. What iscommontoallofthesesuggestionsisthatConvergencecanbeviewedassomethingthat lives on a pre-existing theory of what constitutes candidate phases. This isconceptually more complicated than the view that Convergence exhaustivelydetermines phasehood: convergent objects are phases and nonconvergent objectsare not phases. Anticipating our two approaches to the bound pronoun effectdiscussed below, the approach in section 3.2 (Reining in Convergence) makescrucial reference to phase heads and consequently entails the more complicatedview.Theapproach insection3.2(Reining inTransmission),ontheotherhand, iscompatible with the conceptually simpler view that Convergence exhaustivelydeterminesphasehood.3.1.2TransmissionTheideathatboundpronounscanenterthederivationwithunvaluedphi-featuresgoesbackatleastasfarasKratzer(1998a).(SeealsoHeim,Lasnik&May1991,whoobserve that bound pronouns that range over atomic individuals are nonethelesssometimesmorphologicallyplural due to a syntactic agreement requirementwiththeirantecedent.)Kratzer’s focuswasfakeindexicals, i.e., first-andsecond-personpronounsthatbehaveasboundvariables,likemyinOnlyIfinishedmyhomeworkonthereadingIamtheonlyxsuchthatxfinishedx’shomework.Toaccountforthefactthat the first-person featureonmy seemsnot tobe interpreted,KratzerproposesthatmyentersthederivationwithoutthisfeatureandacquiresitviaagreementinthePFcomponent,afterLFisalreadyfixed.SeealsoKratzer(2009)foranupdated

10

treatment that takes into account, among other things, Rullmann’s (2004) partialbinding data that complicate the basic picture. (An alternative view in the fakeindexicals literature is that the pronoun enters the derivation with the relevantfeaturesbutthesefeaturesgetdeletedpriortoLF:seevonStechow2003.SeealsoReuland2010fordiscussion.)Oneanalyticalchoicepointwithinthisliteraturethatis relevant to our purposes here is the question of whether bound pronounsobligatorilyoroptionallyenterthederivationwithoutphi-features.Heim(2008)forexample argues for the obligatory view, whereas Kratzer (2009) argues for theoptionalview,arguing that long-distance fake indexicalsenter thederivationwithspecified phi-features and have an interpretation that is mediated by context-shifting operators as proposed by Cable (2005). Since the bound pronoun effectunderinvestigationinthispaperissensitivetoasubject/non-subjectasymmetry,itwillbecrucialforusthateither(a)boundpronounsoptionallyenterthederivationwith unvalued phi-features or (b) non-subject “bound pronouns” are not boundpronounsatallbutratherhaveaD-type/E-typeanalysis(cf.Kratzer2009:216).4

Another type of expression that has been argued to involve a pronounenteringthederivationwithunvaluedphi-featuresisPRO.Thissuggestionalsogoesback to Kratzer (1998a, 2009), and has been taken upmore recently by Landau(2015,Toappear).Givenitsrestrictiontosubjectposition,PROinfactbearsamuchcloserresemblancetotheboundpronouneffectthandofakeindexicals.Wereturntothispointinsection5below.3.2ReininginConvergenceThefirstoptionweconsideristoleaveTransmissionasstatedabovebuttomodifyConvergence to account for the subject/non-subject asymmetry. In particular,supposeC isacandidatephasehead,but that its statusasaphasehead isvoidedjustincasetheheadofitscomplementcontainsoneormoreunvaluedfeatures.Thisis spelled out in (21a); (21b) is carried over unchanged from (16b) above. (WethankHisaKitaharaforsuggestingthisapproachtous.)(21)

4Theideathatthird-personpronounscomeintwoflavorsgoesbackatleastasfarasChomsky1955/1975:Chomsky’ssystemhad“twoelementsheandhe*,withhe*a proper noun, and he a pronoun just like I, you” (p. 524 of 1975 edition). ForChomsky, though, the distinction correlated with whether the pronoun had a(sentence-local) antecedent, whereas here, the suggestion is that having anantecedent is a necessary but not a sufficient condition for having entered thederivation with unvalued phi-features. Norbert Hornstein (p.c.) suggests that acomplete theorymayultimatelyneed to countenancenot twobut three flavorsofpronouns:inparticular,thosepronounsthatdonotgiverisetotheboundpronouneffect can be split into two groups depending onwhether they can or cannot bereplacedwithananaphoricepithet.

11

a. Convergence v2: An unvalued feature on the head of a complement to acandidatephaseheadvoidsphasehood.

b. Transmission:Anunvaluedpronouncanbevaluedviafeaturetransmissionfromac-commandingbinderatanarbitrarydistance.

Thisanalysisworksbycapitalizingonthefactthatonepropertythatdistinguishessubjectsfromobjectsandfromsubject-internalconstituentsisthatonlythesubjectparticipates inphi-featureagreementwithT,andTstands ina localhead-to-headrelationshipwithC.Returningtothecrucialdata,thisproposalaccuratelypredictsthat(22)isgeneratedsinceheretheunvaluedstatusoftheboundpronounpreventsembedded T from being valued for phi-features, while also accurately ruling out(23)and(24)sinceinbothcases,theboundpronoundoesnotsitin[Spec,TP]andhencedoesnotpreventtheembeddedTfrombeingvalued.(22)?Joe1claimsthathe1likesapplesandTim2<claimsthathe2likes>oranges.(23)*Joe1claims thatBillgavehim1applesandTim2<claims thatBillgavehim2>

oranges.(24)*Joe1claimsthathis1sonlikesapplesandTim2<claimsthathis2sonlikes>

oranges.

Potentially,thereisapromisingvariantonthisanalysis.SupposePesetsky&Torrego(2001)arecorrectthatembeddeddeclarativeclausesalwaysinvolveeitherT-to-Cmovementormovementofthesubjectto[Spec,CP].Thenwecouldentertainthe ideathatonlyunvaluedfeaturesatthecandidatephaseedge (i.e.,onthephaseheaditselforinthespecifierpositionofthephasehead)voidphasehood:

(25) Convergence v3: An unvalued feature on the edge of a candidate phase

voidsphasehood.InthescenariowhereTmovestoC, theunvaluedfeaturesonTwouldmaketheirway onto C. In the scenario where the subject moves to [Spec,CP], the unvaluedfeaturesonthesubjectpronounitselfwouldnowbeintheedgeofthephase.(Thisproposalwould,however,need tobe refined topreventunvalued features in sub-constituentsoftheelementin[Spec,CP]fromkeepingthephaseopen.Thiswouldbenecessary in order to account for the data involving subject-internal boundpossessors,asin(24).)

ThisvariantoftheproposalinfactoffersaninterestingwayofreinterpretingFelser’s (2004)dataandargumentation.Asdiscussedabove,Felser (2004)arguesthatunvaluedwh-copies in intermediate [Spec,CP]positionskeep thephaseopen,and generalizes from that to suggest that unvalued features in general keep thephase open. But since thewh-copies sit in [Spec,CP], which is at the edge of thephase,itwouldbeconsistentwithFelser’sreasoningtoentertaintheideathatonlyunvaluedfeaturesatthecandidatephaseedgekeepthephaseopen.

12

Taking this line of reasoning even further, (25) offers an interestingalternative perspective onwhat drives successive cyclicmovement: in particular,the suggestion is that successive cyclic movement is not driven by the need toescapetheSpelloutdomaindefinedasthecomplementtothephasehead;rather,itis driven by the need to prevent premature Spellout by voiding the phase headstatusofthecandidatephasehead.

3.3ReininginTransmissionTheotheranalyticaloptiontoconsideristoleaveConvergenceunrestrictedbuttorein inTransmission.Weconsiderhere twovariantsof thisapproach.Onthe firstvariant,transmissionpiggybacksonpredicationandpredicationisdrivenbyaheadthattriggersA-movementsothatthesubject/non-subjectasymmetryfallsoutasasuperiority/minimality effect. On the second variant, transmission piggybacks onbindinginawaythatismediatedbyfunctionalheadsandconstrainedtolocalhead-to-headrelations.

The first variant is inspired by Landau’s (2015) approach to obligatorycontrol,whichamongotherthingsisdesignedtoderivethesubjecthoodofPRO,ananalytical tasksimilar to theoneathandhere. (SeealsoPercus&Sauerland2003for a similar idea in a somewhat different empirical domain.) Suppose, basicallyfollowingLandau(2015), thatphi-feature transmissionpiggybacksonpredication.In(26a)(Landau’s(56a)),forexample,phi-featuresaretransmittedfromsubjecttopredicate, and in pro-drop sentences like (26b) (Landau’s (56b), phi-features aretransmittedfrompredicatetosubject.(26) a. ha-yelad-im smex-im. The-child-PL.M happy-PL.M Hebrew(Landau2015:47) b. pro lleg-ó. he.3.SG arrived-3.SG Spanish(Landau2015:47)Suppose furthermore,still followingLandau(2015), thatacomplementclausecanbe turned into a predicate via a special functional head Fin that has anuninterpretableD feature.This featureattracts theclosestDPto[Spec,FinP].Oncethematrixantecedent ismerged in, itenters intoapredicationrelationwithFinP,enabling feature transmission to the Fin head. As long as subsequent featuretransmissionisconstrainedtoalocalSpec-Headrelationship,thenthesubject/non-subjectasymmetryisderived.Inthelicitderivationin(27)withaboundpronouninsubject position, the pronoun A-moves from [Spec,TP] to [Spec,FinP] (27a),Transmissionoccurs frommatrix antecedent to Fin (27b), and thenTransmissionoccursfromFinto[Spec,FinP](27c).(27) a.Joeclaims[CPthat[FinPheFin[TPhelikesbooks]]]. (A-movement) ↑__________| b.Joeclaims[CPthat[FinPheFin[TPhelikesbooks]]].(Transmission) |______________________________↑

13

c.Joeclaims[CPthat[FinPheFin[TPhelikesbooks]]].(Transmission)↑__|

Non-subject unvalued pronouns, on the other hand, will never lead to a licitderivationwith this setup. The derivation in (28), for example, is ruled out sincemovement of him to [Spec,FinP] incurs a superiority/minimality violation. Thederivation in (29)crashesaswell, since the thirdstep involves transmission fromFintosomethingotherthan[Spec,FinP].(28)Joeclaims[CPthat[FinPhimTP[Billgavehimbooks]]].(*A-movement) ↑__________________|(29) a.Joeclaims[CPthat[FinPBillFinTP[Billgavehimbooks]]].(A-movement) ↑___________| b.Joeclaims[CPthat[FinPBillFinTP[Billgavehimbooks]]].(Transmission)

|_______________________________↑c.Joeclaims[CPthat[FinPBillFinTP[Billgavehimbooks]]].(*Transmission)

|__________________↑In(30),wesummarizetheessentialfeaturesofthisapproach.(30)

a. Convergence:Any unvalued feature anywhere in a candidate phase keepsthephaseopen.

b. Transmissionv2:i. Anunvaluedpronouncanbevaluedviafeaturetransmission.ii. Transmissionofphi-featurespiggybacksonpredication.iii. AcomplementclausecanbeturnedintoapredicateviaFin.iv.Transmission proceeds from antecedent to Fin and from Fin to

[Spec,FinP].

The second variant of this analysis involves taking a Kratzerian view ofbinding and phi-feature transmission: phi-feature transmission piggybacks not onpredication but rather on binding, and binding is mediated by verbal functionalheads.(SeeKratzer1998,2009,drawingonFiner1985;Borer1989;Hale1992.)Inparticular,on thisapproach,amatrixbinder transmits featuresontoembeddedC,andembeddedCbindsandvaluesanunvaluedpronouninitsc-commanddomain.This view derives the subject/non-subject asymmetry, as long as we grant twoadditional assumptions. First, C and v intervene for each other in the way theytransmit features. This ensures that feature transmission can proceed from C to[Spec,TP]asin(31),butitcannotproceedfromCtosomewhereinsidevPasin(32)since thiswould involve an intervening v. The second needed assumption is thatfeaturetransmissionobeysanA-over-A-likeconstraintsothatfeaturetransmissioncannottargetaDPembeddedinalargerDP.Thisthenaccuratelyrulesoutsubject-internalunvaluedpronounsasin(33).

14

(31)Joeclaims[CPthat[TPhelikesbooks]].|_______________↑|_______↑(32)*Joeclaims[CPthat[TPBill[vPgavehimbooks]]]. |______________↑|______________________↑(33)*Joeclaims[CPthat[TP[DP[DPhis]son]likesbooks]].|_______________↑|_______________↑Wesummarizethisvariantoftheanalysisin(34).(34)

a. Convergence:Any unvalued feature anywhere in a candidate phase keepsthephaseopen.

b. Transmissionv3:i. Anunvaluedpronouncanbevaluedviafeaturetransmission.ii. Transmissionofphi-featurespiggybacksonbinding.iii. Bindingismediatedbyverbalfunctionalheads.iv.Candvinterveneforeachotherinthewaytheytransmitfeatures.

4InterventionEffectsIn this section we explore the question of whether there are restrictions on theboundpronouneffectaboveandbeyondtherequirement that theboundpronounbeinsubjectposition.

Onequestiontoaskiswhethertheantecedenttotheboundpronounalsohastobeasubject.Thedata in(35)suggestthattheanswer isyes: inbothsentences,gappingacrossa finite clauseboundary isungrammaticalwhen the subjectof thecomplementclauseisboundbythematrixobject.Thisfollowsfromtheanalysis:anobjectantecedentvalues theboundpronounsufficientlyearly in thederivationsothatthephaseisnotopenlongenoughforcross-clausalgappingtobelicit.(35) a. *Joe1 persuaded Bill2 that he2 should read Pride & Prejudice and Tim3

<persuadedBill2thathe2shouldread>Sense&Sensibility.b.*Joe1promisedBill2 thathe2hadalreadyreadPride&PrejudiceandTim3<promisedBill2thathe2hadalreadyread>Sense&Sensibility.

Notonlydoestheantecedenthavetobeasubject,thedatain(36)showthat

if a matrix object intervenes between the subject antecedent and the boundpronoun, this also blocks the bound pronoun effect. This is also apparent inexampleslike(37)wheretheintervenerisnotamatrixobjectbutratherthesubjectofanintermediateclause.55Interestingly, as shown in (i), when the subject of the intermediate clause is anexpletive, theeffectof the interventionseemstobediminished.Thissuggeststhat

15

(36) a. *Joe1 persuaded Bill2 that he1 should read Pride & Prejudice and Tim3

<persuadedBill2thathe3shouldread>Sense&Sensibility.b.*Joe1promisedBill2 thathe1hadalreadyreadPride&PrejudiceandTim3<promisedBill2thathe3hadalreadyread>Sense&Sensibility.

(37) *Joe1saidthatBillclaimsthathe1likesapplesandTim2<saidthatJoeclaims

thathe2likes>oranges.

Aqualificationisinorderwithregardtothedatareportedhere:sinceeventhebestexamplesoftheboundpronouneffectinvolvedegradedacceptability,it ishardtoknowwhethertotreattheexampleshereascategoricallydifferentinawaythatshouldbebuiltintothegrammarornot.Butsupposetheseinterventioneffectsaregenuine.Whatarewetomakeofthem?Itcouldbethattheyaresimplytellingussomethingabouthowphi-featuretransmissionworks,i.e.,phi-featuretransmissionis subject to intervention. On the other hand, they could be telling us somethingdeeperabouthowtocharacterizetherelationshipbetweentheantecedentandtheboundpronoun.Inparticular,subjectorientationandintervention,takentogether,more or less characterize A-dependencies. Consequently, although we will notpursue this approach here, it is tempting to entertain an analysis in which theantecedent and the bound pronoun constitute anA-chain. Such an approach is infact pursued for copy raising by Potsdam & Runner (2001). Another relevantphenomenon is finite control in Brazilian Portuguese (see e.g. Rodrigues 2004),whichsomescholarshaveanalyzedasinvolvinganA-chaindependencybetweenamatrixargumentandthesubjectofafinitecomplementclause.5ComparisonwithPossiblyRelatedPhenomenaOnemeasureofthesuccessofaproposaloughttobetheextenttowhichitrenderstransparentthetheoreticalrelationshipbetweenseeminglyrelatedphenomena.Asdiscussed above, unvalued pronouns have played a role in the analysis of controland fake indexicals, and so it is worth cataloguing empirical similarities anddifferences between these phenomena and reflecting on how well the analysessketchedabovemayhandlethesesimilaritiesanddifferences. Liketheboundpronouneffect,controlissubject-oriented,PRObeingabletoappear only in subject position (38a) andnot in object position (38b) or subject-

the relevant factor in (37) is the intervening (contentful) DP introduced by theintermediateclauseratherthantheintermediateclauseitself.Thisisreminiscentoftheobservationthatunderatleastsomeconditions,ananaphorneednotbelocallyboundifwhatintervenesisanexpletive(seee.g.Pollard&Sag1994:258—262).(i)*?Joe1saidthat it istruethathe1 likesapplesandTim2<saidthat it istruethathe2likes>oranges.

16

internally(38c).Alsosimilarlytotheboundpronouneffect,thecontrollercannotbeseparatedfromPRObyaninterveningclause,asillustratedin(39).(38) a.Joeclaims[PROtolikeapples].

b.*Joeclaims[(for)BilltolikePRO].c.*Joeclaims[(for)PRO’ssontolikeapples].

(39)Joe1wantedBill2toclaimPRO*1/2tolikeapples.

One empirical difference between the bound pronoun effect and control isthat PRO can be either subject- or object-controlled in a way that famouslycorrelates with the choice of the embedding predicate (40), whereas the boundpronoun effect appears to give rise to ungrammaticality when there is a matrixobject,regardlessofhowtheboundpronounisconstrued,asillustratedin(41).(40) a.Joe1persuadedBill2[PRO*1/2toreadabook].

b.Joe1promisedBill2[PRO1/*2toreadabook].(41) a. *Joe1 persuaded Bill2 that he1/2 had already read Pride & Prejudice and

Tim2<persuadedBillthathe1/2hadalreadyread>Sense&Sensibility.b.*Joe1promisedBill2thathe1/2hadalreadyreadPride&PrejudiceandTim2<promisedBillthathe1/2hadalreadyread>Sense&Sensibility.

Asfaraswecantell,allofthepotentialanalysesoftheboundpronouneffectsketchedinsection3aboveofferapromisingwayofrelatingtheeffecttocontrol.Inessence, we entertained three kinds of explanations for deriving the subjectorientationof theboundpronouneffect: it follows fromeither (1) theneed toget(near) the phase edge, (2) the grammar of predication, or (3) the grammar ofbinding. Deriving properties of control from the grammar of predication and/orbinding isof courseavery familiar idea in the literature (seeLandau2013 foranoverview). As far as we know, there has never been an attempt to derive thesubjecthood of PRO via phase theory (with the partial exception of Landau 2015whoseaccountinvolvesbothphasetheoryandthegrammarofpredication),butweseenoobviousbarriertosuchananalysis. Turning to fake indexicals, here we see some properties that are quitedistinct from the bound pronoun effect. Most strikingly, fake indexicals are notsubject-oriented:allof theexamples in(42)aregrammaticalwitha fake indexicalreading.(42) a.OnlyIclaimedthatIdidthehomework.

Possible reading: ‘I am the only x such that x claimed that x did thehomework.’

b.OnlyIclaimedthatJohnhelpedme.Possiblereading:‘IamtheonlyxsuchthatxclaimedthatJohnhelpedx.

c.OnlyIclaimedthatmyhomeworkwasdone.

17

Possible reading: ‘I am the only x such that x claimed that x’shomeworkwasdone.’

Asfortheinterveningclauseeffect,herethefactsarelessconclusive.Kratzer

(1998:212) generalizes on the basis of sentences like (43a) (repeated from her(15’)) that a fake indexical cannot be separated from its antecedent by aninterveningclause.(43b)alsoseemstosupportthisconclusion.Ontheotherhand,(44),fashionedafterthekindofdatawhoseGermanequivalentwasinvestigatedbyKratzer (2009), seems to permit a fake indexical reading across an interveningclause.(43) a.OnlyIthinkthatMarywon’tcomeifyouinviteme. *’IamtheonlyxsuchthatxthinksMarywon’tcomeifyouwon’tinvitex.

b.OnlyIthinkthatBillsaidJohnsawme.*’IamtheonlyxsuchthatBillsaidJohnsawx.

(44) Iamtheonlyonewhothinksthatsomebodyunderstandsmypaper.

Possible reading: ‘I am the only x such that x thinks that somebodyunderstandsx’spaper.’

Onthebasisof(theGermanequivalentof)datalike(44),Kratzer(2009)proposesthat there are two routes to fake indexicals: pronouns that enter the derivationunvalued and get bound and valued by verbal functional heads in a way that issubject to locality restrictions, and pronouns that enter the derivation valued butwhoseinterpretationismediatedbyCable’s(2005)context-shiftingoperatorsthatarenotsubjecttoanylocalityrequirements.IfKratzer’sproposaliscorrect,thentheempirical differences between the bound pronoun effect and fake indexicals arefully expected, since only the latter can enter the derivation phi-complete andtherebyhavetheoptionofbeinginterpretedviacontext-shiftingoperators.6IslandsIf the basic gist of our analysis is correct— namely that subject-position boundpronouns void the phasal status of the CP— then we make the prediction thatsubject-position bound pronouns should ameliorate certain kinds of islands. Thefollowingminimaltrios involvingadjunct islands(39)andwh-islands(40)suggestthatthispredictionisontherighttrack:ineachcase,nonfiniteclausessignificantlyamelioratetheisland,andamongfiniteclauses,aboundpronouninsubjectpositionrenderstheislandviolationlesssevere.(45) a.What2didJohn1gohome[afterPRO1readingt2]?

b.?What2didJohn1gohome[afterhe1readt2]? c.*What2didJohngohome[afterMaryreadt2]?(46) a.What2didJohn1wonder[whetherPRO1toreadt2]?

b.?What2didJohn1wonder[whetherhe1shouldreadt2]?

18

c.*What2didJohnwonder[whetherBillshouldreadt2]?Inthisconnection, it is interestingtonotethatRoss(1967)questionedChomsky’s(1964) wh-island constraint on the basis that it was too strong, and the datasupportingthispositioninvolvedcontrolledinfinitivalembeddedquestions(47a-d)aswellassixexamplesofembeddedfinitequestionswithbound-pronounsubjects(47e-g,48).6Thedatain(47)-(48)aretakenfromRoss1967:27.(47)HetoldmeaboutabookwhichIcan’tfigureout a.whethertobuyornot. b.howtoread. c.wheretoobtain. d.whattodoabout. e.whyheread. f.?whetherIshouldread. g.??whenIshouldread.(48)Whichbooksdidhetellyou a.whyhewantedtoread? b.?whetherhewantedtoread? c.??whenhewantedtoreadi?In a relatedvein, parasitic gaps arewell known tobebetter innonfinite adjuncts(49a)thaninfiniteadjuncts(49c),anditseemstousthatfiniteadjunctswithboundpronominal subjects pattern with nonfinite adjuncts in being acceptable with aparasiticgap.(49)WhichpapersdidJohnreadbefore… a.…filing? b.…hefiled? c.?...Billfiled?

ButasDavidPesetsky(p.c.)remindsus,itremainsthecasethatextractionofadjuncts out of islands is robustly ungrammatical, regardless of the status of theembeddedsubject,as illustrated in(50).Consequently,we leavefor futureworkamorecompleteinvestigationofboundpronounsinislands.(50) a.*How2didJohn1gohome[afterPRO1solvingtheproblemt2]?

b.*How2didJohn1gohome[afterhe1solvedtheproblemt2]? c.*How2didJohngohome[afterMarysolvedtheproblemt2]?6Interestingly,in(47e),theboundpronounisstructurallyquitefarremovedfromitsantecedent.TheantecedentisthematrixsubjectandtheboundpronounisburiedinsidetherelativeclauseofthematrixPP.Thismaylendsupportfortheapproachtophasehoodandfeaturetransmissionsketchedinsection3.2above,wherebytransmissioncanapplyoveranarbitrarystructuraldistance.

19

7ConclusionThispaperbeganwiththeobservationthataboundpronouninthesubjectpositionofafinitecomplementclauserenderstheclauseboundaryrelativelytransparenttosyntactic processes and relations ordinarily confined tomonoclausal, control, andraisingconfigurations.Weshowedthatthisphenomenonholdsforawiderangeof“quasi-clause-bound” effects including gapping, pseudogapping, comparativedeletion, inverse scope, antecedent-contained deletion, multiple questions, toughmovement,reciprocalbinding,multiplesluicing,andfamilyofquestionsreadings. Towardanexplanation,wesuggestedthattherelevantlocalitydomainforallof these phenomena is the phase, and that bound pronouns have the option ofentering the derivation with unvalued phi-features, thereby voiding phasehood.This basic picture is complicated by the fact that the bound pronounmust be insubject position in order to extend the locality domain, and we entertained twowaysofreiningintheanalysistocapturethis.Onewayinvolvedtheideathatonlyunvaluedfeaturesthatareat(orsufficientlycloseto)thecandidatephaseedgevoidphasehood. The other way involved entertaining the idea that the transmissionmechanismbywhichunvaluedpronounsgetvaluedisconstrainedinwaysthatlimitcross-clausaltransmissiontosubject-positiontargets.Wediscussedwaysofrelatingthisconstrainttothegrammarofpredicationorthegrammarofbinding. Regardlessofthesedetailsofimplementation,though,theoverallgistoftheanalysis has two primary theoretical implications. The first is that not all boundpronouns are created equal: boundpronouns can either enter the derivationphi-complete, or enter the derivation unvalued and thereby interact with coregrammaticalprocesses.ThisconclusionechoesChomsky1955/1975(seenote2)aswell asmore recentwork on bound pronouns such as Kratzer 2009. The secondtheoreticalimplicationisthatnotallCPsarecreatedequal,specificallywithrespectto their phasal status. In summary, the bound pronoun effect offers compellingevidencefortheviewthatfeaturevaluationhasaroletoplayinphasetheory.ReferencesAgüero-Bautista,Calixto.2007.Diagnosingcyclicityinsluicing.LinguisticInquiry

38:413–443. Aissen,Judith,andDavidPerlmutter.1983.Postscripttorepublicationof“clause

reductioninSpanish”.InStudiesinRelationalGrammar,ed.DavidPerlmutter,383–396.Chicago:UniversityofChicagoPress.

Barrie,Michael.2008.Controlandwh-infinitivals.InNewhorizonsintheanalysisofcontrolandraising,ed.WilliamD.DaviesandStanleyDubinsky,263–279.Dordrecht:Springer.

Borer,Hagit.1989.AnaphoricAGR.InThenullsubjectparameter,ed.OsvaldoJaeggliandKennethJ.Safir,69–109.Dordrecht:KluwerAcademicPublishers.

Cable,Seth.2005.Bindinglocalpersonpronounswithoutsemanticallyemptyfeatures.Ms.,MIT.

20

Chomsky,Noam.1964.Currentisusesinlinguistictheory.TheHague:Mouton. Chomsky,Noam.1973.Conditionsontransformations.InAfestschriftforMorris

Halle,ed.S.AndersonandP.Kiparsky,232–286.NewYork:Holt,RinehartandWinston.

Chomsky,Noam.1975.Thelogicalstructureoflinguistictheory.NewYork:Plenum.(Originalworkpublished1955)

Chomsky,Noam.2000.Minimalistinquiries.InStepbystep:EssaysonminimalistsyntaxinhonorofHowardLasnik,ed.RogerMartin,DavidMichaels,andJuanUriagereka,89–155.Cambridge:MITPress.

Chomsky,Noam.2001.Derivationbyphase.InKenHale:Alifeinlanguage,ed.MichaelKenstowicz,1–52.Cambridge:MITPress.

Cinque,Guglielmo.2004.‘Restructuring’andfunctionalstructure.InStructuresandbeyond:Thecartographyofsyntacticstructures,ed.A.Belletti,volume3,132–191.NewYork:OxfordUniversityPress.

Farkas,DonkaF.,andAnastasiaGiannakidou.1996.Howclause-boundedisthescopeofuniversals?InSALTVI,ed.TeresaGallowayandJustinSpence,35–52.Ithaca,NY:CornellUniversity.

Felser,Claudia.2004.Wh-copying,phases,andsuccessivecyclicity.Lingua114:543–574.

Finer,Daniel.1985.Thesyntaxofswitch-reference.LinguisticInquiry16:35–55.Frank,RobertandDennisStoroshenko.2015.Experiencingscope:Inverted

expectationsofQRinraising.PaperpresentedatLinguisticSocietyofAmericanAnnualMeeting,Portland,OR,1/8/15.

Fox,Danny.2000.Economyandsemanticinterpretation.Cambridge:MITPress.Goodall,Grant.1991.Wanna-contractionasrestructuring.InInterdisciplinary

approachestolanguage:EssaysinhonorofS.-Y.Kuroda,ed.C.GeorgopoulosandR.Ishihara,239–254.Dordrecht:Kluwer.

Grano,Thomas.2012.Controlandrestructuringatthesyntax-semanticsinterface.PhDdissertation,UniversityofChicago.

Grano,Thomas.2015.Controlandrestructuring.Oxford:OxfordUniversityPress.Hale,Kenneth.1992.Subjectobviation,switchreference,andcontrol.InControland

grammar,ed.R.K.Larson,S.Iatridou,U.LahiriandJ.Higginbotham,51–77.Dordrecht:KluwerAcademicPublishers.

Heim,Irene.2008.Featuresonboundpronouns.InPhitheory:Phi-featuresacrossmodulesandinterfaces,ed.DanielHarbour,DavidAdger,andSusanaBéjar,35–56.Oxford:OxfordUniversityPress.

Heim,Irene,HowardLasnik,andRobertMay.1991.Reciprocityandplurality.LinguisticInquiry22:63-101.

Higginbotham,James.1981.Reciprocalinterpretation.JournalofLinguisticResearch1:97–117.

Hornstein,Norbert.1994.Anargumentforminimalism:Thecaseofantecedent-containeddeletion.LinguisticInquiry25:455–480.

Johnson,Kyle.1996.InsearchoftheEnglishmiddlefield.Ms.,UniversityofMassachusetts,Amherst.

Kayne,RichardS.1998.Overtvs.covertmovement.Syntax1:128–191.

21

Kennedy,Christopher.1997.Antecedent-containeddeletionandthesyntaxofquantification.LinguisticInquiry28:662–688.

Kratzer,Angelika.1998a.Morestructuralanalogiesbetweenpronounsandtenses.InProceedingsofSemanticsandLinguisticTheoryVIII,ed.DevronStrolovitchandAaronLawson,92–109.CornellUniversity:CLCPublications.

Kratzer,Angelika.1998b.Scopeorpseudoscope?Aretherewide-scopeindefinites?InEventsandGrammar,ed.SusanRothstein,163–196.KluwerAcademicPublishers.

Kratzer,Angelika.2009.Makingapronoun:Fakeindexicalsaswindowsintothepropertiesofpronouns.LinguisticInquiry40:187–237.

Kuno,SusumuandJaneJ.Robinson.MultipleWhQuestions.LinguisticInquiry3:463-487.

Landau,Idan.Toappear.AgreementatPF:AnargumentfromPartialControl.Syntax.

Landau,Idan.2015.ATwo-TieredTheoryofControl.Cambridge:MITPress. Larson,RichardK.,andRobertMay.1990.Antecedentcontainmentorvacuous

movement:ReplytoBaltin.LinguisticInquiry21:103–122. Lasnik,Howard.2002.Clause-mateconditionsrevisited.GlotInternational6:94–96. Lasnik,Howard.2006.Afamilyofquestions.Handout,USC.Lasnik,Howard.2014.MultiplesluicinginEnglish?Syntax17:1–20.Lechner,Winfried.2001.Reducedandphrasalcomparatives.NaturalLanguage&

LinguisticTheory19:683–735.Lee-Schoenfeld,Vera.2007.Beyondcoherence:ThesyntaxofopacityinGerman.

Amsterdam:JohnBenjaminsPublishingCompany.May,Robert.1977.Thegrammarofquantification.PhDdissertation,MIT.May,Robert.1985.Logicalform:Itsstructureandderivation.Cambridge:MITPress.Merchant,Jason.2001.Thesyntaxofsilence:Sluicing,islands,andthetheoryof

ellipsis.Oxford:OxfordUniversityPress.Moulton,Keir.2008.Clausalcomplementationandthewager-class.InProceedingsof

NELS38,ed.A.SchardlandM.Walkow,165–178.Amherst,MA:GLSAPublications.

Nishigauchi,Taisuke.1998.‘multiplesluicing’inJapaneseandthefunctionalnatureofwh-phrases.JournalofEastAsianLinguistics7:121–152.

Percus,OrinandUliSauerland.2003.Pronounmovementindreamreports.InProceedingsofNELS33,ed.M.KadowakiandS.Kawahara,265–284.Amherst:GLSAPublications.

Pesetsky,DavidandEstherTorrego.2001.T-to-Cmovement:Causesandconsequences.InKenHale:Alifeinlanguage,ed.MichaelKenstowicz,355–426.Cambridge:MITPress.

Pollard,CarlandIvanSag.1994.Head-DrivePhraseStructureGrammar.Chicago:UniversityofChicagoPress.

Postal,Paul.1974.Onraising:OneruleofEnglishgrammaranditstheoreticalimplications.Cambridge,MA:MITPress.

Potsdam,EricandJeffreyT.Runner.2001.Richardreturns:Copyraisinganditsimplications.InProceedingsfomtheMainSessionoftheThirty-seventhMeetingof

22

theChicagoLinguisticSociety,ed.MAndronis,C.Ball,H.Elston,andS.Neuvel,453-468.Chicago:ChicagoLinguisticSociety.

Reuland,Eric.2010.Minimalversusnotsominimalpronouns:Featuretransmission,featuredeletionandtheroleofeconomyinthelanguagesystem.InTheLinguisticsEnterprise:Fromknowledgeoflanguagetoknowledgeinlinguistics,ed.M.B.H.Everaert,T.Lentz,H.DeMulder,Ø.Nilsen,andA.Zondervan,257–282.Amsterdam:JohnBenjamins.

Rizzi,Luigi.1978.ArestructuringruleinItaliansyntax.InRecenttransformationalstudiesinEuropeanlanguages,ed.SamuelJ.Keyser,113–158.Cambridge:MITPress.

Rodrigues,Cilene.2004.ImpoverishedmorphologyandA-movementoutofcasedomains.PhDdissertation,UniversityofMaryland.

Ross,JohnR.1967.Constraintsonvariablesinsyntax.PhDdissertation,MIT.Rullmann,Hotze.2004.Firstandseconpersonpronounsasboundvariables.

LinguisticInquiry35:159–168.Ruys,E.G.1992.Thescopeofindefinites.PhDdissertation,UtrechtUniversity.Sloan,Kelly.1991.Quantifier-whinteraction.MITWorkingPapersinLinguistics

15:219–237.Sloan,Kelly,andJuanUriagereka.1998.Whatdoes‘everyone’havescopeover?

GLOW.Budapest.vonStechow,Arnim.2003.Featuredeletionundersemanticbinding.InProceedings

ofNorthEastLinguisticSociety(NELS)33,ed.MakotoKadowakiandShigetoKawahara,377–403Amherst:UniversityofMassachusetts,GLSA.

Williams,Edwin.1986.AreassignmentofthefunctionofLF.LinguisticInquiry17:265–299.

Wurmbrand,Susi.2001.Infinitives:Restructuringandclausestructure.Berlin: MoutondeGruyter.

Wurmbrand,Susi.2011.Onagreeandmerge.RevisedcoursenotesfromProblemsinSyntax(Spring2011),UniversityofConnecticut,accessed5/17/13.

Wurmbrand,Susi.2013.QRandselection:Covertevidenceforphasehood.InProceedingsoftheNorthEasternLinguisticsSocietyAnnualMeeting42,ed.StefanKeineandShayneSloggett,277–290.Amherst:UniversityofMassachusetts,GLSA.

Wurmbrand,Susi.2015.Restructuringcross-linguistically.ToappearinProceedingsofNELS45.

(Grano)DepartmentofLinguisticsIndianaUniversitytgrano@indiana.edu(Lasnik)DepartmentofLinguisticsUniversityofMarylandlasnik@umd.edu